There are two major components to mechanistic modeling: model parameterization and model validation using empirical data. Each model should be parameterized and validated at many locations to ensure that the plant parameters accurately reflect plant growth across space and time. Model parameterization for a site can be performed using detailed field data collected for functionally important plant traits or by gleaning reasonable values from previous findings in the literature. These functional traits may be morphological (i. e., leaf area), physiological (i. e., stomatal conductance or proportion of N in tissue) and phenological (i. e., date of green-up, senescence, and maximum growing degree days) characteristics. Model parameterization often requires significantly more data collection over time than model validation. Therefore, many studies do not independently parameterize each model but only change key variables known to vary between locations (Kiniry et al. 2008).

Model validation, comparing measured yields to simulated yields for many sites is becoming increasingly common as more field trials managed for biofuel production are being performed (Wullscheleger et al. 2010). A review by Wullschleger et al. (2010) revealed 17 switchgrass studies managed for biofuel production at 39 different locations. Management varies widely across these studies and can greatly impact biomass production. For example, yearly N fertilizer application in these studies ranges from 0 to 896 kg N ha1. Changes in management practices need to be carefully considered when validating model output using yields from several studies.

To our knowledge, there are currently 14 studies that have used mechanistic models of switchgrass growth to predict biomass production. The number of times each model has been parameterized and validated for switchgrass varies (Table 1). The ALMANAC model has been most extensively parameterized and validated with five studies parameterizing the model and six studies validating the estimated yields.